
Mbodi is working with companions comparable to ABB Robotics. Supply: Mbodi AI
Producers are dealing with worsening labor shortages, however programming typical robots to answer adjustments in manufacturing can take weeks. Consequently, billions of {dollars} in high-mix, low-volume manufacturing stays out of attain for automation, in keeping with Mbodi AI.
The New York-based firm in the present day introduced that it has been chosen to hitch the Y Combinator accelerator’s X25 batch.
“We’re constructing an embodied AI platform that lets anybody train robots new abilities — simply by speaking to them. No code. No engineers. No week-long reprograms,” said Mbodi AI co-founders Xavier (Tianhao) Chi and Sebastian Peralta. “Mbodi turns pure language and fast demos into exact, dependable robotic actions that run in manufacturing in minutes, and adapt as typically as you want.”
Chi and Peralta, who helped run Google Public DNA (8.8.8.8) at Google, based Mbodi AI final yr. In December, ABB named the New York-based firm as a winner of its 2024 ABB Robotics AI Startup Problem.
As well as, Mbodi has been acknowledged with a 2025 RBR50 Robotics Innovation Award. Chi spoke in a panel on generative AI’s impression on robotics on the Robotics Summit & Expo earlier this month.

Xavier Chi and Sebastian Peralta have been engineers at Google earlier than co-founding Mbodi. Supply: Mbodi AI
Automation wants flexibility for adoption
Handbook labor drives half of gross home product (GDP), or $40 trillion, however 70% of factories worldwide face labor shortages, in keeping with Mbodi AI. Why hasn’t automation unfold extra rapidly?
“It’s fascinating — once I received into the sector, I discovered that the world is lots much less automated than I assumed,” Chi replied to The Robotic Report. “For example, individuals suppose packing is generally automated, however lots continues to be executed by people.”
Reprogramming robots requires on-site specialists and time, he mentioned. Producers can’t sustain with ever-shifting manufacturing wants and rising demand, in order that they want a neater solution to instruct robots.
“Most duties require some degree of flexibility, and automation often includes a protracted improvement cycle,” mentioned Chi. “On the manufacturing flooring, individuals don’t have the experience to run these robots. Collaborative robots have been initially designed to be bought straight, however finish prospects should have a tough time with integration and price, that are limitations to mass deployment.”
Now accepting session submissions!
Mbodi AI helps robots study duties
Mbodi mentioned its cloud-to-edge system combines generative AI, agent orchestration, and symbolic reasoning to show high-level directions into exact, protected, and adaptive robotic actions. The startup claimed that its “common agent framework can run throughout any robotic {hardware}, coordinating notion, reasoning, planning, and management in actual time.”
The corporate sees agentic AI as invisible to the person, who simply tells the robotic what to do, mentioned Chi. One agent may get information from the world, whereas one other may break down duties or spawn new brokers.
“We just like the scene in Iron Man the place the robotic simply follows directions and passes Tony Stark instruments,” Chi recalled. “It was fiction earlier than, nevertheless it’s coming to life now, due to generative AI.”
“Presently, most robots must be programmed to do a particular factor, so that you want integrators,” he added. “That has been the case since these robots have been invented many years in the past. However now, you possibly can repurpose a robotic extra simply in the present day and inform it what to do.”
Mbodi AI’s interface abstracts a number of the management layer, bringing it along with imaginative and prescient for object detection and path planning, for a extra generalized intelligence, defined Chi.
Chi additionally famous that when one robotic learns a job, it will probably share that data with your complete fleet. This permits particular person abilities to be reused, repurposed, and tailored throughout machines, factories, and duties.
AI robotics startup seeks companions
Mbodi AI mentioned it’s rolling out manufacturing deployments to prospects with assist from companions comparable to ABB Robotics. This may deliver its product into real-world amenities to automate duties that have been beforehand too advanced or pricey to program.
The corporate mentioned its objective is “a world the place robots are as simple to show and adapt as software program, turning them into true client commodities, and the place automation and robotics turn into accessible to each manufacturing facility and past.”
As a part of its Y Combinator announcement, Mbodi is seeking to join with producers, methods integrators, and robotic makers to resolve issues that have been beforehand out of attain for robotics.
The corporate is initially specializing in pick-and-place purposes. Chi mentioned he sees potential for cell manipulation and better ranges of autonomy.
“There will likely be particular smaller imitation fashions, which is able to want an clever system layer to coordinate them. The totally different fashions will likely be task-specific or based mostly on an organization’s personal information set,” he mentioned. “We see our system transferring towards expertise-specific VLA [vision-language-action] fashions.